The era of ‘Intelligence +’ has quietly arrived, and advanced science and technology such as AI technology and machine learning have begun to penetrate into the construction industry, bringing new design concepts. In this paper, based on building information modelling and machine learning, an automatic matching method for indoor building style layout is designed. Morphological parameters are extracted according to the designed indoor architectural form, and the mapping relationship between spatial form and architectural style is established based on the GAN model. An orthogonal model is used to model the generated architectural style image, and the spatial relationship of the indoor building, as the premise of the GAN model for style matching. It is found that traditional indoor style layout matching takes about 10 minutes, while the automated matching method proposed in this paper takes only 18.5 seconds, which greatly improves the matching efficiency. Even under different constraints, the method is still effective in matching interior architectural design solutions that simultaneously meet the contour constraints, as well as the style constraints.
